A Unified Network Selection Framework Using Principal Component Analysis and Multi Attribute Decision Making

摘要

Services are ubiquitously delivered over multiple wireless access technologies in a heterogeneous wireless network environment. A significant issue is the ranking of the alternative access networks and the selection of the most efficient and suitable one in order to meet the Quality of Service (QoS) requirements of a specific service, as these are defined by the user. With this way, the user receives enhanced Quality of Experience. However, decisions on which network to connect are usually difficult to be reached, since multiple factors of different importance have to be taken into consideration. The subject of the paper is the introduction of a novel unified network selection framework. A framework for defining decision criteria weights relying on the variance of network measurements is also provided. For this purpose, Principal Component Analysis and Analytic Hierarchy Process are deployed so as to unravel the patterns in data and retrieve parameter weights through pair wise comparisons respectively. The framework addresses the final network selection by the employment of Multi Attribute Decision Making methods and by using certain QoS indicators acting as figures of merit which influence the decision process. Finally, the proposed scheme is tested through extended simulations and a discussion over its performance is made. Some useful conclusions are drawn.